Add shader rendering

[dbierek]

This PR improves the speed of rendering by using:

• Frustum culling
• Occlusion culling with asynchronous depth sorting
• Backface culling
• Batched drawing, where all copies of each aloc (or prim) are drawn in a single gl draw call
• Shaders with VBOs and VAOs instead of immediate mode

It also loads the Obj to Recast and to the renderer in parallel.